I also am talking about "trust" as in "trustworthy", not the security technical definition. I think we're saying the same thing, but I lay the blame on an inherent aspect of the system, not on the Google/MS/Facebook big players in the space.
Any site owner (be it Google or Mom's BBQ Shack) cannot accept third party authentication, without implicitly relying on whatever user creation policies that third party uses to control their audience.
If tomorrow Google suddenly opened the floodgates and said spambots could create all the Google IDs they wanted, then practically overnight you would see wholesale disabling of Google ID authentication on sites that currently use it.
The reality is that no-one other than the really big players get enough public attention to be considered trustworthy for 3rd party authentication. Allowing unrestricted third-party authentication services by definition means allowing anonymous accounts. And truly anonymous accounts are diametrically opposite from having logged-in users.
My point is that this isn't a Google/big data tracking/hate the corps issue. The point of user logins are to provide you (the site owner) controls over your userbase. If you offload your logins to 3rd parties, you are sacrificing most (if not all) of those controls.
Here's a real example - I run a site that has a private area. Users are authenticated using Facebook (because I don't want to force extra logins on them). It's cut down on the vast majority of bogus signup attempts, but only because Facebook is relatively good about preventing spambots from creating accounts. But there's no way in hell I would allow Mom's BBQ Shack to provide authentication (aka, OpenID) because I have no visibility or public evaluation on how Mom's BBQ Shack creates logins. For all I know, Mom's BBQ Shack is really just a spam king, and I just allowed spambot logins on my system.
We have a couple of great examples of truly anonymous, distributed systems, where every node is equal allowed behavior: Email and Usenet. Spam problems on both are fundamentally insolvable without breaking the systems to rely on outside methods of trust. The same applies for an authentication service. You cannot have a fully open and anonymous system, without it allowing for anonymized abuse.